SPARSE CONVOLUTIONAL NEURAL NETWORK ACCELERATOR

专利名称：SPARSE CONVOLUTIONAL NEURAL

NETWORK ACCELERATOR

发明人：William J. Dally,Angshuman Parashar,Joel

Springer Emer,Stephen William Keckler,LarryRobert Dennison

申请号：US158837申请日：20170314

公开号：US20180046916A1公开日：20180215

专利附图：

摘要：A method, computer program product, and system perform computations

using a sparse convolutional neural network accelerator. Compressed-sparse data isreceived for input to a processing element, wherein the compressed-sparse data encodesnon-zero elements and corresponding multi-dimensional positions. The non-zeroelements are processed in parallel by the processing element to produce a plurality ofresult values. The corresponding multi-dimensional positions are processed in parallel bythe processing element to produce destination addresses for each result value in theplurality of result values. Each result value is transmitted to a destination accumulatorassociated with the destination address for the result value.

申请人：NVIDIA Corporation

地址：Santa Clara CA US

国籍：US

更多信息请下载全文后查看